The statistics of Premier League football have been stood on their head by a season that has started with too few draws, the FT reports today.

 Can there be too few draws? For bookmakers there can. Punters prefer to bet on wins, so the bookies do better if there are more draws. This season, report Simon Kuper and Roger Blitz, there have so far been only four draws in 66 matches. On past experience, we might have expected at least 16.
 
Is this a blip, or a trend? Kuper quotes Stefan Szymanski, a sports economist at Cass Business School in London, as saying that the change is statistically significant. But is it meaningful?
 
Kuper cites two reasons for believing that it might represent a real change. One is that the Premiership is increasingly dominated by five teams so much stronger than the rest that they normally win easily. Their wealth has provided them with powerful strikers who score regularly – the Premier League is averaging 2.97 goals a game, the most for the top division of English football for 41 years.
 
Here's the graph, based on data from www.statto.com
 
              
 
The second reason, Kuper says, is that managers have worked out that striving to win is a better policy than settling for a draw, even if the strategy occasionally comes unstuck. A win has provided three points since 1981, against just one for a draw, so if he is right it has taken them a long time to work this out.
 
However, it makes sense. If a club that is drawing with 20 minutes to go and takes risks in order to win, the strategy ought to pay off. If, in three games, it turns one draw into a win, fails to alter the score in another, and by taking risks loses the third, it will still have garnered four points, against three for settling for a draw in all three games.
 
But this has been true here since 1981, and is true everywhere since football worldwide adopted “three points for a win” in 1995. Yet in continental Europe draws are running at the usual level this season. Does this imply managers there are stupider than managers here? The opposite is normally believed to be the case.
 
To my eye, the odd statistics in this season’s Premier League look like a classic example of bunching. It is unexpected for a football league to generate so few draws, but take in all the leagues in the world and it’s not surprising that one of them does. It happens to be ours, so we look for reasons.
 
Chances are there aren’t any, and that the ratio of draws will regress towards the mean. We’re not even a fifth of the way through the season yet. If I’m right, betting on draws for the rest of the season ought to be profitable. If I’m wrong, then football really has changed and the bookies will have to rethink their odds.